test(ocr): add business-logic tests for polygon extraction, Kraken routing, and confidence markers

Cover Surya polygon/word-level extraction, health endpoint states,
Kraken script-type routing, 503 when models not ready, 400 when
Kraken unavailable for Kurrent, and confidence marker application
during streaming. Production code coverage: 88%.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

ocr-service/test_engines.py

@@ -45,6 +45,52 @@ def test_surya_extract_page_blocks_returns_blocks_for_single_image():
     assert blocks[0]["height"] == 20 / 200
 
 
+def test_surya_extract_page_blocks_extracts_polygon_when_present():
+    image = Image.new("RGB", (100, 200))
+
+    mock_line = _make_surya_line("Text", [10, 20, 90, 40])
+    mock_line.polygon = [(10, 20), (90, 20), (90, 40), (10, 40)]
+
+    mock_pred = MagicMock()
+    mock_pred.text_lines = [mock_line]
+
+    with patch.object(surya, "_recognition_predictor") as mock_rec, \
+         patch.object(surya, "_loaded", True):
+        mock_rec.return_value = [mock_pred]
+
+        blocks = surya.extract_page_blocks(image, page_idx=1, language="de")
+
+    assert blocks[0]["polygon"] is not None
+    assert len(blocks[0]["polygon"]) == 4
+    assert blocks[0]["polygon"][0] == [10 / 100, 20 / 200]
+
+
+def test_surya_extract_page_blocks_extracts_word_level_confidence():
+    image = Image.new("RGB", (100, 200))
+
+    word1 = MagicMock()
+    word1.text = "Hallo"
+    word1.confidence = 0.95
+    word2 = MagicMock()
+    word2.text = "Welt"
+    word2.confidence = 0.3
+
+    mock_line = _make_surya_line("Hallo Welt", [10, 20, 90, 40], words=[word1, word2])
+
+    mock_pred = MagicMock()
+    mock_pred.text_lines = [mock_line]
+
+    with patch.object(surya, "_recognition_predictor") as mock_rec, \
+         patch.object(surya, "_loaded", True):
+        mock_rec.return_value = [mock_pred]
+
+        blocks = surya.extract_page_blocks(image, page_idx=1, language="de")
+
+    assert len(blocks[0]["words"]) == 2
+    assert blocks[0]["words"][0]["text"] == "Hallo"
+    assert blocks[0]["words"][0]["confidence"] == 0.95
+
+
 def test_surya_extract_blocks_delegates_to_extract_page_blocks():
     """After refactoring, extract_blocks should produce the same output."""
     image1 = Image.new("RGB", (100, 200))

ocr-service/test_stream.py

@@ -161,3 +161,111 @@ async def test_old_ocr_endpoint_still_returns_flat_list(mock_images):
         assert isinstance(data, list)
         assert len(data) == 2
         assert data[0]["pageNumber"] == 1
+
+
+# ─── Health and error handling ────────────────────────────────────────────────
+
+
+@pytest.mark.asyncio
+async def test_health_returns_ok_when_models_ready():
+    with patch("main._models_ready", True), \
+         patch("main.kraken_engine") as mock_kraken:
+        mock_kraken.is_available.return_value = True
+
+        async with AsyncClient(transport=ASGITransport(app=app), base_url="http://test") as client:
+            response = await client.get("/health")
+
+        assert response.status_code == 200
+        data = response.json()
+        assert data["status"] == "ok"
+        assert data["surya"] is True
+        assert data["kraken"] is True
+
+
+@pytest.mark.asyncio
+async def test_health_returns_503_when_models_not_ready():
+    with patch("main._models_ready", False):
+        async with AsyncClient(transport=ASGITransport(app=app), base_url="http://test") as client:
+            response = await client.get("/health")
+
+        assert response.status_code == 503
+
+
+@pytest.mark.asyncio
+async def test_ocr_stream_returns_503_when_models_not_ready():
+    with patch("main._models_ready", False):
+        async with AsyncClient(transport=ASGITransport(app=app), base_url="http://test") as client:
+            response = await client.post("/ocr/stream", json={
+                "pdfUrl": "http://minio/test.pdf",
+                "scriptType": "TYPEWRITER",
+            })
+
+        assert response.status_code == 503
+
+
+@pytest.mark.asyncio
+async def test_ocr_stream_uses_kraken_for_kurrent_script(mock_images):
+    with patch("main._download_and_convert_pdf", new_callable=AsyncMock, return_value=mock_images), \
+         patch("main._models_ready", True), \
+         patch("main.kraken_engine") as mock_kraken:
+        mock_kraken.is_available.return_value = True
+        mock_kraken.extract_page_blocks.return_value = [_make_block(1, "Kurrent text")]
+
+        async with AsyncClient(transport=ASGITransport(app=app), base_url="http://test") as client:
+            response = await client.post("/ocr/stream", json={
+                "pdfUrl": "http://minio/test.pdf",
+                "scriptType": "HANDWRITING_KURRENT",
+            })
+
+        lines = [json.loads(line) for line in response.text.strip().split("\n") if line.strip()]
+        page_events = [l for l in lines if l["type"] == "page"]
+        assert len(page_events) == 3
+        mock_kraken.extract_page_blocks.assert_called()
+
+
+@pytest.mark.asyncio
+async def test_ocr_stream_returns_400_when_kraken_unavailable_for_kurrent(mock_images):
+    with patch("main._download_and_convert_pdf", new_callable=AsyncMock, return_value=mock_images), \
+         patch("main._models_ready", True), \
+         patch("main.kraken_engine") as mock_kraken:
+        mock_kraken.is_available.return_value = False
+
+        async with AsyncClient(transport=ASGITransport(app=app), base_url="http://test") as client:
+            response = await client.post("/ocr/stream", json={
+                "pdfUrl": "http://minio/test.pdf",
+                "scriptType": "HANDWRITING_KURRENT",
+            })
+
+        assert response.status_code == 400
+
+
+@pytest.mark.asyncio
+async def test_ocr_stream_applies_confidence_markers(mock_images):
+    """Low-confidence words should be replaced with [unleserlich] in the stream output."""
+    def page_blocks(image, page_idx, language="de"):
+        return [{
+            "pageNumber": page_idx,
+            "x": 0.1, "y": 0.2, "width": 0.8, "height": 0.1,
+            "polygon": None,
+            "text": "original text",
+            "words": [
+                {"text": "Lieber", "confidence": 0.95},
+                {"text": "xkqz", "confidence": 0.1},
+            ],
+        }]
+
+    with patch("main._download_and_convert_pdf", new_callable=AsyncMock, return_value=mock_images[:1]), \
+         patch("main._models_ready", True), \
+         patch("main.surya_engine") as mock_surya:
+        mock_surya.extract_page_blocks.side_effect = page_blocks
+
+        async with AsyncClient(transport=ASGITransport(app=app), base_url="http://test") as client:
+            response = await client.post("/ocr/stream", json={
+                "pdfUrl": "http://minio/test.pdf",
+                "scriptType": "TYPEWRITER",
+            })
+
+    lines = [json.loads(line) for line in response.text.strip().split("\n") if line.strip()]
+    page = [l for l in lines if l["type"] == "page"][0]
+    assert page["blocks"][0]["text"] == "Lieber [unleserlich]"
+    assert "words" not in page["blocks"][0]